OPENING ACCESS, ENDING THE VIOLENCE TRAP:

LABOR, BUSINESS, GOVERNMENT AND

THE NATIONAL LABOR RELATIONS ACT

MARGARET LEVI, STANFORD UNIVERSITY

TANIA MELO, UNIVERSITY OF WASHINGTON

BARRY WEINGAST, STANFORD UNIVERSITY

FRANCES ZLOTNICK, STANFORD UNIVERSITY

SEPTEMBER 29, 2016

Abstract

Open access to labor organizations lagged nearly a century behind open access to business

organizations, arising as part of the New Deal in the mid-1930s. During this century, firms and

governments actively suppressed labor organization, frequently resorting to large-scale violence.

All this ended with the National Labor Relations Act (NLRA) of 1935. Why did the violence

associated with labor last for a century? What did the NLRA do to solve this problem, and why

couldn’t Congress have done so earlier? The purpose of this paper is to develop a new

perspective on labor organization and violence that addresses these questions. We argue that the

century-long violence surrounding labor resulted from an inability to solve a series of

commitment problems. All three parties to the violence – labor, business, and government –

faced commitment problems. We show that the NLRA succeeded because it finally solved the

commitment problems underlying the century of labor violence.

1

1. Introduction

Throughout American history, open access has been contested. Those in power used

violence to deny various groups the right to organize effectively or to otherwise make political

claims. In their attempt to date open access in the United States, North, Wallis, and Weingast

(NWW) (2009) focused on general incorporation. This criterion dates open access in the United

States in the 1840s following a series of general incorporation laws that open access to business

organizations.

We raise two problems with the NWW account of the emergence of open access in the

United States. First, widespread open access in the United States was more fraught and

contested than NWW suggest, and it certainly took longer to achieve than they originally

realized. Open access to labor organization, for example, would take a century longer than open

access for business. The claims of African-Americans for full citizenship were ignored or

repressed for a century after emancipation, and the explicitly racial violence against them was

too long tolerated. This aspect of open access remains a struggle to this day. Second, demands

for open access by excluded groups evoked violence by government, private actors, and

claimants.

In this paper, we extend the NWW argument about the emergence of open access where

that access was violently contested. In particular, we follow North and Weingast (1989) to argue

that the solution of the violence problem required inventing a new set of credible commitments.

This paper extends the understanding of how difficult credible commitments are to achieve and

sustain even in seemingly democratic civil societies.

2

Some excluded groups made insistent claims for open access, most notably labor

organizations. Yet, for the century prior to the New Deal of the 1930s labor was denied the

organizational rights and privileges.1 During the century previous to the New Deal, firms and

governments actively suppressed labor organization, firing workers who struck, blacklisting and

arresting labor leaders, and deporting those who were immigrants.

The United States during this 100-year long period failed two critical conditions

articulated by NWW as requisite for open access: first and most obviously, the denial of open

access to labor; and second, the use of violence to suppress labor. At least until the end of the

turmoil of the 1930s, many firms employed private armies to protect them from strikes and

unionization. Ford’s “Service Department,” composed of underworld thugs and mercenaries,

was infamous for its violence and intimidation tactics (Bernstein 1971, 735-751). Governments

used the police, National Guard, and the U.S. Army at times to crush nascent labor organization,

leading frequently to killings and mass beatings.

Labor peace emerged only with the invention of new administrative structure and

process; that is, a set of new regulatory institutions that solved a series of commitment problems

that plagued the emergence of non-violent resolution of disputes. These new institutional

solutions to the commitment problem arose in the New Deal with the passage of the National

Labor Relations Act (NLRA) and the creation of the National Labor Relations Board (NLRB) in

1935. Since the late 1930s, labor violence has been far lower and labor-firm cooperation far

1 Other excluded groups acquiesced in the face of violent repression, notably African Americans. Robert Dahl,

one of the great champions of pluralist democracy, effectively argued that where there was no noise, there was no

claim (Dahl 1961). The civil rights movement convinced him that sometimes, as his critics Bachrach and Baratz

(1962, 1963; also see Lukes 2005 [1974]) argued, smoldering anger and frustration lay behind quiescence.

3

higher. In the words of Taft and Ross (1969:292), “The sharp decline in the level of industrial

violence is one of the greatest achievements of the National Labor Relations Board.”

This short sketch of labor history raises several questions that remain unanswered in the

literature. Why was the labor violence so intractable? What exactly did the NLRA/NLRB do that

– somehow – solved the problem of violence? And, if this legislation solved the problem, why

didn’t Congress do so earlier, thereby saving the deadweight losses associated with years of

violence, strikes, and a considerably lower level of cooperation between firms and their workers?

Although substantial gains from cooperation existed among the three major groups, all

three faced commitment problems. Business – fearful of labor’s threat to its control over

business management, the labor force, and to corporate profits – could not commit to eschew

violence. Nor could government commit to being an impersonal arbiter instead of being an agent

of firms against labor. Too often, government officials associated labor organization with

anarchy and revolution, and it considered business a source of stability and economic growth.

Further, the law of property and contracts favored business, providing an important legal basis

for government to collaborate with firms. Labor could not commit to eschewing political

demands for foundational changes in the economy; nor could it commit extremists to forego

violence at moments when the great majority would prefer not to.

The stakes were therefore high. Legalization of unions would foster the growth of

powerful actors in opposition to business, making labor demands more pressing. Without solving

labor’s commitment problems, business was rationally reluctant to support legislation that would

authorize unions. The result was on-going violent suppression of labor with considerable

foregone gains from cooperation between labor and business.

4

The 1930s legislation channeled labor-business conflict to focus on wages and working

conditions, an outcome that was not pre-ordained. Much of the literature implicitly accepts these

bounds by ignoring the central problem of violence. So why and how was solution

institutionalized in the NLRA? Motivating the change was labor’s existential threat to business

during this period when unions and labor organization were perceived as potential collaborators

in a growing radical, even revolutionary, movement in the United States (Katznelson 2013). We

further argue that the acceptance and sustaining of the legislation also required transformations

in the substance and implementation of administrative law.

The NLRA had several well-known accomplishments. It legalized unions, required

collective bargaining; it defined a number of common anti-union tactics as “unfair labor

practices” and hence illegal; and it created an enforcement mechanism to make it work.

In addition, however, the legislation accomplished several ends largely unrecognized in

the literature. We list three. First, the NLRA dramatically lowered the stakes for firms. It

narrowed considerably the legitimate range of bargaining between labor and business, focusing

on wages and conditions; the legislation removed labor’s threat to business management and

firm capital; it also prevented unauthorized strikes, helping unions control their more radical and

extreme elements who favored goals beyond wages and benefits.

Second, the legislation transformed government from an advocate of business using

violence against labor into an impersonal arbiter, impersonal in the sense that regulators had

incentives to punish either side for failing to abide by the rules.2 Equally important, the

legislation provided obvious advantages for labor. It legitimized unions, allowing labor

2 Although the NLRB became an impersonal arbiter, the legislation was nonetheless heavily biased in favor of

organized labor.

5

organization to form, grow, and advance workers’ interests. As union ranks grew considerably,

labor became an important political force, able to support its position in a manner not previously

possible. By counterbalancing business, labor provided new and substantive support for the

NLRA an impersonal arbiter.

Third, to accomplish these ends, organizational and legal innovations were necessary to

create a new form of regulatory delegation that sat comfortably within the constitutional

framework. Put simply, for the new system to work, political officials and the courts had to solve

the principal agency problem that we now take for granted: creating a regulatory agency that

implements the intentions of Congress while not transgressing the due process rights of citizens

and firms.

Our framework affords answers to each of the questions we asked at the outset. Labor

violence proved long-lived and intractable because of commitment problems, None of the three

parties – labor, business, and government – were willing or capable of unilaterally eschewing

violence. The NLRA ended a century of violence because it solved the various commitment

problems facing the three sets of players. Finally, this legislation could not have been

implemented earlier because it required significant innovation in public law and organization that

occurred only in the context of the multi-pronged regulatory framework of the New Deal.

Labor unions, more than most other civil society organizations, were perceived as

threatening; and given the various the problem of violence reflecting difficult commitment

problems, this perception was rational.

This paper proceeds as follows. Section 2 provides a brief history of the labor movement

up to the early 1930s, including the legal context that, prior to 1930, favored business over labor,

6

leading the government to protect property and hence to side with business. Section 3 provides a

simple game theoretic analysis showing why, prior to the mid-1930s, violence remained a central

component of the equilibrium. Section 4 considers the historical and institution shifts that

ultimately produced a relatively non-violent equilibrium under the NLRA and NLRB. In section

5, we adapt the game presented in section 3 to incorporate the new institutions created by the

NLRA; this analysis reveals how the NLRA solved the various commitment problems facing the

three many actors, business, government, and labor. Section 6 develops a new interpretation of

the constitutional controversies of the 1930s that led to solving the commitment problems. In

section 7, we present our implications and conclusions.

2. A Brief History of the Labor Movement

The barriers to organizing were high and remained insurmountable for most workers in

the United States until the 1930s.3 The most important obstacles were the unremitting hostility of

the managers, the alliance of the government with the corporations, and the diversity of interests,

demands, and organizational strategies within the labor movement itself. None of the parties

could credibly commit to refrain from violence. Management believed it was their right to

defend their prerogatives and profit with private armies and physical intimidation. Government

believed corporate interests trumped those of labor and regularly brought in the National Guard

to settle strikes on behalf of industry owners. Labor had no means to control those in their ranks

who chose violent confrontation.

3 For accounts of the history of American labor unions in the nineteenth and early twentieth century, see

Montgomery (1989), (Hattam 1993), (Perlman 1928; Voss 1993), and Foner (1947). For accounts that extend into

the early 1940s, see Bernstein (1969, 1971, 1985) and Lichtenstein (2002). Also, see Brecher (1997) for a

provocative history of strikes in the United States.

7

Numerous accounts contrast the business ideology in the United States with other

countries from at least the nineteenth century on. American business had a long-standing

opposition to unionization that distinguished it—well into the twentieth century—from many

European countries (Friedman 1988; Swenson 2002; Mares 2003). The federal government

generally shared that opposition except for particular moments in time, such as World War I

when government depended on the unions sufficiently to overcome its antagonism and act as a

mediator between business and labor. Strikes over time in the United States reveal that periods in

which government engaged in mediation or impersonal intervention in strikes helped resolve the

violence problem while also facilitating effective bargains and unions, those that were able to

compromise and promote productivity (Geraghty and Wiseman 2011).

The particular history of the relations among business, government, and labor affected

union organizing and legislative strategies with consequences for the timing of the achievement

of open access. The rise in labor militancy in the mid to late nineteenth century set the stage for

the practices and laws that would influence access until the 1930s. Currie and Ferrie (Currie and

Ferrie 2000), for example, explore heterogeneous state-level legal frameworks that emerged as

workers engaged in disputes to win collective bargaining and how those laws affected

subsequent strike activity. Using data from 1881-1894, they document the relationship between a

higher incidence of labor disputes when legislation, most notably injunctions, reduced worker

certainty about strike outcomes; they report a lower incidence of disputes where legislation, most

notably protective laws such as maximum hours, increased worker certainty.

Other research focuses on the organization strategies of unions. Friedman (1988), for

example, compares France with the U.S. and argues that the more hostile government attitude in

the United States led unions to avoid or prevented them from relying on the state. The American

8

Federation of Labor (AFL), the first successful long-lived union confederation, embodied the

approach to disputes that Friedman outlines. It was built on craft workers, rather than industrial

workers. The crafts unions were the first to succeed at legal organization through their control of

jobs and accreditation, achieving the acceptance of employers and government by the end of the

nineteenth century. These unions monopolized the supply of craft labor through an

apprenticeship system and hiring halls, effectively requiring employers to come to the unions and

pay the union-set rates. The founding of the American Federation of Labor (AFL) in 1886

further consolidated the craft workers’ legal and political position but at the expense of the

unskilled workers and the newer immigrants (Perlman 1928).

AFL victories derived largely from agreements with employers in a given industry; the

Federation seldom appealed to government to set minimum wages or maximum hours. Its self-

stated mantra was “voluntarism” – that is, “relying on their own voluntary organizations…

defended the autonomy of the craft union against the coercive intervention of the state” (Rogin

1962, 521-2). The Federation lobbied but most determinedly to reduce job competition; among

its major campaigns was immigration restriction. During the 1930s, for example, the AFL

expressed cautious support of social security and considerable nervousness about all legislation,

including the National Labor Relations Act and other legislation that might interfere with its

internal affairs (Eidlin 2009, 253).

By contrast, industrial unions had a far more difficult time establishing themselves.

Employers and governments often met large-scale strike waves, generally coinciding with the

cycle of depressions that began in the mid-nineteenth century, with violence. Industrial unions

also experienced numerous stops and starts. The Knights of Labor was the first large-scale labor

American organization; it experienced a period of rapid growth after its founding in 1869 but

9

was dead by the 1890s (see Voss 1993 and Foner 1947). Labor activism intensified at the end of

the nineteenth century and again in the teens of the twentieth century, but the same pattern

prevailed: worker mobilization followed by repression.

The rise of the Industrial Workers of the World (IWW) or “Wobblies” created a new kind

of political threat. The IWW professed revolutionary goals, and it believed in direct action over

political action (cf Adler 2011, Kimeldorf 1969). It used strikes to disrupt the economy, not just

to improve working conditions. A militant union that organized all workers, craft and industrial,

IWW’s presence was particularly strong among miners, lumberjacks, and dockworkers. Its name

became associated with violence in the public mind when its leadership, most notably “Big Bill”

Haywood, who was also on the executive committee of the U.S. Socialist Party, was accused of

masterminding the assassination of Governor Frank Steunenberg of Idaho in 1905, presumably

in retaliation for putting down an 1899 miners’ strike the governor had labeled an “insurrection.”

The prosecution was secretly bankrolled by the mine owners, but Clarence Darrow’s defense led

to the acquittal of the accused (Lukas 1997). Haywood was also among the hundred or so IWW

members convicted in 1918 under the Espionage Act of 1917, but he escaped prison by fleeing to

the Soviet Union where he lived his remaining years.

The “Red Scare” of 1919 closed a chapter in American labor history. While “the Great

War” raged in Europe, but also in its immediate aftermath, the United States experienced

multiple major strikes, considerable labor organizing, violent confrontations between police and

unions, terrorist acts (including bombings and assassinations) by revolutionary anarchists, and

Socialist electoral victories. Some of these actions were illegal and violent; others, such as the

Seattle General Strike of 1919 (Johnson 2008) and the Boston Police Strike of the same year

(Levi 1977), were peaceful but illegal; and some, such as the electoral strategy of Eugene V.

10

Debs and the Socialist Party, were non-violent and legal. But all ultimately got tarred with the

same brush, as fear of mayhem and revolution became widespread among the public.

Although President Woodrow Wilson proclaimed his support of labor during the

Versailles discussions (Lichtenstein 2002, 4), the United States simultaneously attempted to rid

the labor movement of its militant leadership. The Espionage Act of 1917 and the Immigration

Act of 1918 increased the power of the federal government to deport any persons it deemed

dangerous to the national interest. The “Palmer Raids” of 1919 that bear the name of the attorney

general Wilson appointed, permitted jailing of leaders and members of radical organizations and

the closing down of their offices. Newspapers, business leaders and government officials fueled

“the rationality of fear” (deFigueiredo and Weingast 1999), and Americans, as a rule, feared the

violent revolution they had come to believe was possible if its perpetrators were not repressed.

Fear of a revolutionary labor movement resurfaced in the 1930s but with significant

differences. The Great Depression created a large pool of dispossessed, unemployed, and

disgruntled citizens. The Communist Party offered an alternative vision of the future with

promises of economic security and equity that the present United States did not seem capable of

delivering—economically or politically. Although committed in principle to the violent

overthrow of the United States, the Communist Party did not use violence, was legal until the

1940s, and worked with and through numerous other organizations, including unions.

Deportation was no longer an effective weapon given that almost all the workers – militant social

reformers, and Communists –were American-born citizens. However, employers and

governments still used repression. The Minneapolis Teamster strike of 1934 exemplifies the

times: A more radical local was put down by a combination of the international union and the

federal government, which jailed some of the strike leaders in part for being Trotskyists

11

(Ahlquist and Levi 2013).

Large-scale industry, which had begun in the late nineteenth century, was steadily

expanding and automobile manufacture had become one of the biggest. The assembly line that

began with the Ford Motor Company put workers side by side in huge factories. The assembly

lines were dehumanizing, but they also gave workers new power to disrupt production.

The new industrial unions, too, differed from those that had preceded them. John L.

Lewis, head of the mineworkers, first proposed in 1928 what was to become the Congress of

Industrial Organizations (CIO). Although not actually established until 1935, the strategies and

ideologies of the CIO’s leaders were the dominant influence on the big strikes of the 1930s. The

leaders were, as rule, committed social democrats; even those affiliated with the Communist

Party lacked the revolutionary fervor of earlier radical leadership; and some, such as Lewis

himself, were relatively conservative politically and strongly anti-communist (Bernstein 1969,

126). Leaders focused on organizing all the workers within their factories and industries, be they

skilled or unskilled, but also on forming effective alliances across industries. And they were

willing to engage in large-scale strikes.

The legal context

Although some unions gained the right to exist in some states and for some occupations

in the early nineteenth century, industrial unions did not gain full associational rights until the

1930s.4 If one follows the four-fold distinction in Brooks and Guinnane, both the rights to

assemble (R1) and to form a “mere association” (R2) were legally problematic and often

4 Even with the passage the National Labor Relations Act (NLRA) in 1935, many unions, most notably those

among agricultural or domestic workers, did not receive full associational rights. Moreover, within a decade of the

passage of the NLRA, concerted opposition to open access by unions began to succeed.

12

physically repressed throughout the nineteenth century (see Bloch and Lamoreaux). Obtaining

legal recognition as an “associational entity” (R3) was granted unevenly, with some craft unions

firmly winning that right by the end of the nineteenth century, industrial unions only in the

1930s, and public employees and farm workers not until the last part of the twentieth century.

Full incorporation (R4) as a legal person or entity was increasingly permitted but often rejected

by unions, which resisted state intervention in their internal affairs and attempted to evade the

possibilities of suits against the labor organization itself.5 Moreover, the early bases for

incorporation depended on qualifying as a mutual aid society or charity. Legal incorporation has

little to do with and often inhibited the essential characteristics of unions: collective bargaining

and the right to strike. For labor organizations, open access means the combination of legal

recognition of their right to exist, bargain, and strike.

Prior to the New Deal, the law systematically favored employers against labor. The

police powers were designed to protect life and property. In general, "Even where the police

were not directly suborned by employers, their primary duty was the defense of the employer's

property, and in this sense they participated in industrial disputes as partisans. The very presence

of the police or troops at a struck plant carried with it the implication that the strikers were

lawbreakers. It signified that strikers were the enemies of public order, for quite obviously the

police had not been summoned to protect them, but company property from them" (Gitelman

1973:17).

5 Although unincorporated, unions could still be subject to suit, however. In United Mine Workers of America v.

Coronado Coal Co., 42 S.Ct. 570 (1922), at 576, the court found "In this state of federal legislation, we think that

such organizations are suable in the federal courts for their acts, and that funds accumulated to be expended in

conducting strikes are subject to execution in suits for torts committed by such unions in strikes."

13

General incorporation laws worked positively to support business organization and to

further a range of legitimate business purposes, such as their right to use the courts to protect

their interests. The same did not hold for labor. In the late 19th and early 20th centuries, "The

common law legality of unionism, however, did not confer a right to organize. It merely left

workingmen free to form unions when and if they could" (Gitelman 1973:6). The absence of a

legitimate way to organize meant that striking workers were often seen as a mob potentially

threatening employer property, which the law was designed to protect.

The absence of legislation legitimizing labor and, especially, labor organization, had a

series of implications. While labor sought to interpret strikes and walkouts as temporary

absences, many firms interpreted strikes as a permanent disruption of employment, making it

legal for them to hire new employees (seen by workers as strikebreakers). As Gitelman explains

(1973:9), "The expectation of returning to work at the conclusion of a strike was jeopardized by

the legal and popularly sanctioned right of employers to hire and fire at will."

Moreover, the legal system in the late 19th

and early 20th

centuries tolerated what we

would today think of as “unorthodox” – indeed, illegitimate – means of resisting labor’s attempts

to organize and bargain with firms. Firms used armed men against strikers, fired labor

organizers, and dismissed workers for joining unions; firms also refused to listen to workers'

complaints or grievances, suppressed worker free speech, and widely used spies and agent

provocateurs who, for example, sought to incite workers to use violence and even initiated

violence.

The government and employers also used legal tools against labor. These included

antitrust laws used against labor organizations, as discussed in section 5. Injunctions became a

14

staple used to prevent strikes and reduce labor’s leverage (Frankfurter and Greene 1930). Forbath

(1991) presents the following table of estimated labor injunctions over time:

Table 2.1. Labor Injunctions

By Decade, 1880-1929.

Decade Injunctions

1880s 105

1890s 410

1900s 850

1910s 835

1920s 2130

Source: Forbath (1991, Appendix B)

The table provides evidence of the frequency with which employers and governments used this

legal tool to suppress labor activity and organization.

During the late nineteenth and early twentieth century, the AFL led a battle against the

use of injunctions and other judicial devices for restraining labor’s associational power (Tomlins

1985, 61-68). Increasingly, an argument emerged over the legal personhood of the union.

Without it, there was a judicial and legislative question whether the unions could be held

accountable for their members’ actions and could, therefore, contract on their behalf. Indeed,

“…as early as 1895, arguments were held in Congress advocating the treatment of unions in law

as organizations possessed of the capacity to bind their members. By doing so, it was held,

unions could be enlisted as disciplinary agents in the service of labor-management peace”

(Tomlins 1985, 84).

The debate about the appropriate legal status of unions was at the heart of the discussion

of the Erdman Railway Arbitration Act in the 1890s, a bill that represented an effort to give labor

15

organizations legal personality (Tomlins 1985, 84-86). The Railway Brotherhood supported the

bill, but the AFL opposed it on the grounds that it was against the interest of unions to be held

accountable for members.

In the years immediately preceding World War I and during the war years themselves,

the law became more tolerant of unions although far from recognizing full associational rights.

Since the passage of the Sherman Anti-Trust Act in 1890, the courts had defined labor

organizations as monopolies; this act was the basis of most of the injunction activity. The

Clayton Act of 1914 exempted labor organizations from the monopoly and cartel provisions of

the Sherman Act and, in principal, limited the used of injunctions to circumstances where

property was threatened. During World War I, the dependence of business on its increasingly

scarce labor led to an enhancement of union rights and benefits in practice if not always in law.

With the end of the WWI and the rise of anti-radicalism embodied in Attorney General

Palmer’s raids, anti-unionism again became the norm. The law continued to favor employers,

placing a large number of varied constraints against labor organization. The passage of the

Railway Act of 1926 was the first significant sign of thaw, but it was not until the New Deal that

labor organizations more generally gained open access.

3. The Labor Organization Game, 1880-1930.

From the late 19th

century through the early 1930s, the United States faced a “violence

trap” (Cox, North and Weingast 2014) with respect to labor. None of the key players – the

government, labor, and business – had the ability to commit not to use violence. Although we

assume that each player preferred legalization, negotiation, and no violence to violence,

commitment problems prevented the three players from obtaining this outcome.

16

To understand the commitment problems, we model the union (U), business (B), and

government (G) interaction as a 3 player game.6 We use variants on a game to represent three

different periods: between roughly 1880-1930; 1933-36, and 1937-forward. In focusing on these

three players, we abstract from differences among unions, among businesses, and within the

government. In doing so, we gain greater analytic power to derive important implications about

labor violence and some of the major mechanisms that helped solve the problem of violence.

Here we outline the initial game. We shall discuss each of the subsequent games

following the periods of history that affected them.

LABOR ORGANIZATION GAME Period 1: 1880-1930

The sequence of play in the first game (1880-1930) is as follows. U has the initial move

and must choose from three choices (see figure 1). First, U can strike within limits; meaning that

U avoids violence and that U limits its demands to wages and working conditions. Second, U can

strike without limits, possibly using violence and possibly demanding more from employers than

just wages and better conditions (such as representation in management or a seat on the corporate

board). Finally, U can choose to revolt, possibly leading to a better political compromise for

labor but, more likely, to disorder, repression, and large scale violence. Specifically, if U chooses

this option, nature (N), a non-strategic player, chooses between two outcomes: with probability

p1, N chooses A, in which U wins an attractive political compromise; and with probability (1 –

p1), N chooses outcome C. The subscript on p indicates the period. We define L31 as the implied

lottery following U choice of revolt, where the first subscript indicates the lottery number (in this

6 This game follows in the tradition established by Golden’s (1997) analysis of labor institutions that explains why

strikes in some countries are far more severe than in others.

17

case, 3) and the second subscript indicates the time period (1 for 1880-1930, 2 for 1933-36, and 3

for 1937-forward); thus, L31 = p1A + (1 – p1)C.

If U chooses to strike (either within or without limits), B has the next move and must

choose between responding with violence or by negotiating with U.

Finally, consider G’s decisions between siding with business and doing nothing. If U

strikes (with or without limits) and G sides with B to fight U, N chooses an outcome. If U has

chosen to strike within limits, then with probability q1, N chooses outcome D representing

attractive concessions from B to U; and with probability (1 – q1), N chooses outcome E, resulting

in violence against workers, destruction of the union organization, and no concessions. These

three choices by U, B, and G yield lottery L1 = q1D + (1 – q1)E. If, in contrast, U chooses to

strike without limits, then N chooses between D and E, but the probability of D (concessions

from B) is q2 while the probability of E is (1 – q2). The implied lottery of N’s choice is L2 = q1D

+ (1 – q1)E. We assume that q2 > q1; that is, U has greater leverage when it ignores limits on its

actions and hence U is more likely to prevail if it does not adhere to limits on its demands.

Based on the history presented above, the players’ preferences are as follows. First, U

prefers D (concessions) to E (destruction of the union). Second, it prefers L2 to L1. To see this,

notice that if U chooses to strike within limits, the lottery L1 ensues; if strike without limits, the

lottery L2 ensues. The only difference between the two lotteries is the probability of the

concessions to labor. Because q2 > q1, U prefers L2 over L1; technically, we have L2 =q2D + (1 –

q2)E > q1D + (1 – q1)E =L1. Third, given the relatively low likelihood (p1) of a successful revolt

during this period, U prefers L2 to L31.

Next consider B’s preferences. If U strikes (with or without limits), B is best off fighting.

The reason is that B prefers L1 to I and L2 to F. Because U cannot commit to refrain from

18

violence, negotiation means that B gives concessions, while U cannot commit to honoring the

concessions it makes.

Finally, consider G’s preferences. The period under consideration, 1880-1930, is a

largely Republican era in which that party typically held the presidency and a majority in both

houses. Republicans favored business generally and, particularly, the protection of property

rights and freedom of contract. Both union organization and labor violence threatened business

through restriction of property rights and freedom of contract. Hence, during this era, G prefers

to support business and use violence to suppress labor. Specifically, G prefers L1 to outcome I;

and L2 to outcome F.

We can now solve for the equilibrium of the game using backward induction. Consider

the first terminal node in the upper right; U has chosen to strike within limits and B has chosen to

fight. Because G prefers L1 to I, G chooses to side with B. Next, consider the second terminal

node. If U chooses to strike without limits and B chooses to fight, then G also chooses to side

with B (G prefers L2 to F). Working backwards one node to B’s decisions. If U has chosen to

strike (with or without limits), B chooses to fight since it prefers L1 to J and L2 to H. Finally,

consider U’s choice at the initial node of the game. We have established that U prefers L2 to L1;

and because p1 is low, it prefers L2 to L31. Hence U will choose to strike without limits. Along the

equilibrium path, then, U chooses to strike without limits; B chooses to fight; and G sides with B

against U.

The equilibrium of this game represents an on-going and sometimes violent conflict

between business and labor, with the government siding with business against labor. Labor

strikes without limits, and hence violent confrontations between business and labor are on-going,

and the government actively attempts to suppress labor organization and bargaining with firms.

19

The passage of the Wagner Act changed the incentives of the players and, therefore, the

equilibrium of the game.

4. NLRA and NLRB

Congress passed the National Labor Relations Act (NLRA), also known as the Wagner

Act, in 1935 in part to stem a rising tide of industrial violence of the 1930s. The National

Industrial Recovery Act (NIRA) of 1933, intended to foster economic recovery, inadvertently

spurred unrest by providing symbolic support for worker organization, but without the

institutional machinery necessary to implement and protect that right. The NIRA encouraged a

major organizing drive by labor unions, but a lack of enforcement encouraged employers to

resist. Increasing disparity between labor’s de jure and de facto rights led to unprecedented

levels of industrial conflict, which impeded the already fragile economic recovery.

The NIRA suspended antitrust law and permitted industry and trade associations to

formulate codes of competition that would regulate production within industries. Recognizing

that allowing economic combination of firms would greatly advantage business relative to

workers in the labor market, language was added to strengthen the position of labor. Section 7a

required all industry codes to meet three conditions: provide a right for employees to organize

and bargain collectively through representatives selected without interference from their

employer, prohibit compulsory membership in company unions, and require employers to

comply with minimum wage rates, maximum hour limitations, and other regulations on working

conditions as approved by the President.

Labor interpreted Section 7a as a call to organize and launched an unprecedented

organizing drive in 1933. Strike activity, which had declined to historic lows in the 1920s,

20

surged. As both economic activity and organizational drives increased, man-days lost to work

stoppages jumped, from fewer than 603,000 monthly in the first half of 1933 to 1,375,000 in July

and 2,378,000 in August, threatening the fragile economic recovery (Bernstein 1950, 58).

Empowering an Impersonal Arbiter

Perhaps the most important accomplishment of the NLRA was its provision of the Board

with the structure and powers necessary to enforce the constraints on both sides. The explosion

in industrial unrest after the passage of NIRA stemmed largely from the disconnect between the

rights promised to labor under section 7(a) and those actually realized. This disconnect resulted

from a virtually complete lack of enforcement (Gross 1974, 129).

Section 3 of the NLRA empowered the NLRB, a quasi-judicial, independent agency of

the federal government, to oversee union elections, certify representatives, and investigate and

prevent unfair labor practices. The design of an effective NLRB involved two different

processes. First, a process of learning, in part through trial and error; and second, a more careful

study of the regulatory and administrative structure and process previously sanctioned by the

Supreme Court, such as the Federal Trade Commission (1914) and the Interstate Commerce

Commission (1887), as Irons (1983) emphasizes.

Between 1933 and 1935, the Roosevelt administration, officials from the NRA and the

Department of Labor, and interested parties from Congress experimented with the agency

structure. The first National Labor Board (NLB) was established in August of 1933, to

adjudicate, mediate, and conciliate disputes arising under section 7(a) of NIRA. But the board

had limited investigatory powers, and relied upon other agencies, often with conflicting interests,

for enforcement. At the request of NLB chairman and Senator Robert Wagner, Roosevelt issued

21

a series of Executive Orders to strengthen the Board by reducing the level of review that other

agencies had over NLB decisions and giving the Board authority to oversee representation

elections. Yet the Board still had no enforcement power, and overlapping jurisdictions between

the NLB and the NRA led to contradictory statements of policy and encouraged employers to

ignore the Board’s orders. Reliance on the Department of Justice for access to the courts caused

further bureaucratic and administrative problems (see Tomlins 1985, 109-119).

Following an unsuccessful attempt to pass a precursor the NLRA, Congress instead

passed Public Resolution 44, which replaced the NLB with a new board called the National

Labor Relations Board (hereafter the Old NLRB). This Board had a non-partisan structure,

greater investigatory powers, and more independence through non-reviewable findings of fact.

Yet, the Old Board’s election orders were reviewable by circuit courts, which allowed employers

to delay elections through court challenges, and the Board still relied upon other agencies for

enforcement of its decisions.

The final structure and process of the NLRB under the NLRA solved a number of the

problems that had plagued prior agencies and that had undermined their effectiveness. The most

important innovations concerned three aspects of the Board’s power: (1) strengthening

investigatory powers, allowing the Board to amass suffient evidence; (2) providing the Board

with exclusive jurisdiction so that other agencies with conflicting goals could not block the

board; and (3) removing barriers to enforcement of Board decisions. Like the definition of unfair

labor practices and the election and bargaining unit provisions, these aspects of the Board’s

structure were in part the product of prior boards’ members’ experiences in battling with

employers to enforce section 7(a) of the NIRA.

22

Section 11 of the NLRA outlined the Board’s investigatory powers, providing it with the

authority to subpoena witnesses and evidence, the ability to appeal to District Courts for

enforcement of subpoenas, and requiring that other government agencies provide information

upon request. Section 12 provided for substantial penalties, including up to 1 year of jail time,

for interference with or resistance to Board investigations. Prior boards’ lack of subpoena power

had two adverse affects: it allowed employers to impede investigations by simply ignoring Board

requests to testify; and it impeded enforcement, as the pre-NLRA boards had to rely on either the

National Recovery Administration or the Department of Justice (DOJ) for enforcement; the DOJ

required cases referred by the Board to be complete in all legal details before it would accept

them, and since the Board had no subpoena powers, it could not meet this requirement. Sections

11 and 12 of the NLRA solved these problems.

Section 10(e) allowed the Board to bypass these middlemen entirely and petition the

circuit courts for enforcement of orders directly; it also made the Board’s findings of fact

conclusive. Enforcement through the DOJ and NRA had been ineffective for reasons greater than

the Board’s lack of evidence gathering powers. The NRA’s only enforcement tool was to rescind

business’ Blue Eagles, the license allowing them to operate under NIRA industry codes. In

practice, removal of Blue Eagles had little effect, and the agency was disinclined to do so, since

its own mandate was to promote economic activity rather than inhibit it. Similarly, the DOJ was

unenthusiastic about enforcing board orders, and initiated all proceedings de novo, due to the

weakness of the boards’ own investigatory powers. Board members thought the DOJ staff to be

“unsympathetic, lax, and in many cases incompetent (Gross 1974, pg. 129)”, and the duplicative

investigations slowed down enforcement to the point of total nullification: between July 1934

23

and March 1935, for example, no judgments were obtained in any of the 33 NIRA

noncompliance cases referred to the DOJ by the board (Bernstein 1950, 87).

Direct appeal to the courts increased the autonomy of the Board by removing the

effective, if informal, veto power that the NRA and the DOJ exercised over the Board’s

judgments. Section 10(a) gave the NLRB an exclusive right to prevent unfair labor practices. The

law’s authors sought to make the NLRB the “supreme court” of labor, to prevent the confused

jurisdiction over labor disputes that had arisen under NIRA as the board, the NRA, and

Roosevelt himself all sought independently to solve labor disputes arising under Section 7(a).

Without exclusive jurisdiction, the board’s decisions were frequently undermined by

contradictory statements of policy from the NRA. Roosevelt often got involved in negotiations to

try to bring major work stoppages to an end, and to that end carved whole industries out of the

early boards’ jurisdiction in order to give them dispensations from board principles.

To summarize our argument: the NIRA asserted various labor rights to organize, but

failed to create an effective set of administrative structure and process to enforce them:

NIRA provided no clear mandate, command structure, or process to create rules and

precedents with which to regulate union activity and labor-firm bargaining. For example,

it failed to define adequately the type of acceptable organizations designed to represent

union members; and it created no process, structures, or substance by which a firm could

be found not in compliance with the law.

Unclear lines of authority created a range of bureaucratic and administrative problems:

The law required that the NLB rely on the NRA and DOJ for enforcement, each of which

had their own priorities that conflicted with those of the NLB.

President Roosevelt intervened in ad hoc ways inconsistent with the NRA.

The constitutional status of the law and hence NLB regulations remained uncertain,

affording employers the ability to delay and resist NLB authority.

In the face of this confusion, the absence of clear constitutionality, and the inability of the

government to enforce the rules, employers resisted at every turn. The disparity between promise

and actuality in the context of the Depression generated unprecedented labor unrest.

24

What did the NLRA do?

The NLRA resolved each of these problems. It granted the NLRB a clear mandate with

substantially more effective mandate and effective structure and process. The act clarified lines

of authority. It also gave the Board the direct ability to enforce its rulings without relying on

other organizations, including subpoena powers. By making the NLRB the sole legal authority in

its area, the Act also removed the ability of the president to intervene within the agency’s

jurisdiction. In stark contrast to the 1933 legislation, the act was consciously designed to

maximize the likelihood that the Supreme Court would find it constitutional. Finally, the

Supreme Court’s acceptance of the NLRA’s constitutionality led to enforcement of the act,

employer compliance, and an end to violence associated with labor.

Broadly, the NLRA accomplished two ends. First, It asserted a federal right for workers

to organize and bargain collectively, via a representative of their own choosing. Second, section

3 established the National Labor Relations Board (NLRB), an independent, quasi-judicial agency

to adjudicate disputes arising under the law. Yet neither of these institutions originated in the

NLRA. As noted, the right to organize was first asserted in NIRA (the language of section 7 of

the NLRA was drawn directly from section 7a of the NIRA), and precursors to the NLRB had

existed since August of 1933. These earlier measures failed, however, to stem the violence

problem that pervaded labor relations in the 1930s and earlier. How, then, did the NLRA differ?

The NLRA succeeded where prior attempts had failed because it went beyond earlier

legislation in five ways: (1) It defined a number of unfair labor practices that by nature interfered

with the meaningful enjoyment of the organizing and bargaining rights created in the law,

imposing clear and uncontestable constraints on employers; (2) it provided a Board-controlled

process for election of representatives, effectively constraining labor as well; (3) it provided the

25

NLRB with the power and independence necessary for effective enforcement of those constraints

upon both workers and their employers; (4) it cleared up lines of authority so that, first: the

president could not intervene on an ad hoc basis; and second the NLRB did not depend, as did its

predecessors, on other organizations for enforcement; and, (5) it created a regulatory process that

the Supreme Court held constitutional and hence legally binding on employers.

The last two accomplishments are the basis for the development of administrative law

that transforms the right of open access into a reality.

Constraints on Employers

The unfair labor practices defined in Section 8 of the NLRA provided explicit statutory

support for NLRB prosecution of one general and four specific employer practices that

undermined workers’ right to organize, hold recognition elections, and bargain collectively. To

provide the Board with the flexibility to address practices not anticipated during the writing of

the legislation, section 8(1) included a blanket prohibition on “interference with, restrain[t], or

coerc[ion]” of employees in their exercise of rights guaranteed in section 7. Sections 8(2)

through 8(4) banned employer dominated (company) unions, discrimination of any sort to

encourage or discourage membership in unions, and discrimination or retaliation against workers

who testified or filed charges under the NLRA. Section 8(5) addressed the most common and

disruptive reason for labor conflict during this period, by making the refusal to bargain

collectively with elected representatives an unfair labor practice.

The definition of unfair labor practices provided, for the first time, a statutory basis for

NLRB intervention in a set of employer practices that undermined workers’ stated right to

organize. The NLRB’s predecessor boards had established precedents for such intervention, but

26

the lack of a clear legislative mandate and contradictory statements by the National Recovery

Administration (NRA) and Roosevelt had encouraged employers to challenge or ignore these

decisions.

Constraints on Workers

The 1935 NLRA defined only employer-side unfair labor practices; prohibitions on

union-side activities would be added more than a decade later in the 1947 Taft-Hartley

amendments. On its face, New Deal labor policy thus appears to impose limits on employers

without constraining the behavior of unions. We argue, however, that the law provided

meaningful limits on both employers and unions. The election process and rules defined in

section 9 of the NLRA provided a standardized process for acquiring the benefits of NLRA-

protected collective bargaining. This process, and the gatekeeping role of the NLRB in certifying

the outcome, effectively constrained the behavior of workers and their unions as well as

employers.

Section 9 of the NLRA established the rules and procedures for the election of bargaining

representatives, and the role of the NLRB in this process. First, section 9(a) codified two

important principles that had been the source of many legal challenges to bargaining: majority-

rule elections and an exclusive right to representation for the winner of such an election. Under

NIRA, employers had challenged the results of union elections by arguing that majority-rule

elections deprived the minority of the right to be represented by an organization of their choice.

Both employers and unions, when they lost an election, asserted that representatives preferred by

minority groups should have standing to bargain as well.

27

The authors of the NLRA, many of whom were members of the NLRB’s predecessor

boards, feared that the fracturing of bargaining authority would undermine the goal of collective

bargaining and exacerbate the problem of inter-labor disputes. Bernstein writes: “The experience

of the Auto Board [an industry-specific labor board that had allowed for multiple

representatives] convinced the draftsmen that pluralism provoked confusion and strife, defeating

collective bargaining (1950, 96).” Section 9(a) of the Act therefore declared that

“[r]epresentatives designated or selected for the purposes of collective bargaining by the majority

of the employees in a unit appropriate for such purposes, shall be the exclusive representatives of

all the employees in such unit for the purposes of collective bargaining[.]”

Second, section 9(b) gave the NLRB the right to define the scope of the bargaining unit.

This issue was the source of significant opposition from the AFL, since it moved decision-

making power on a critical strategic problem to regulators. Elections take place within the

bargaining unit; therefore, the definition of the unit has the potential to make- or- break unions.

These decisions are highly strategic – unions want the unit to be big enough to have leverage

against the employer, but election campaigns are easier to manage within a smaller unit.

Because the selection of the bargaining unit size and composition has important implications for

the relative strength of employer versus union, the drafters placed this power in an impersonal

party: the NLRB. Bernstein writes, “[T]hey sought to avoid placing the authority in the

employer, which might invite violations of the act, and to employees, who might use it to defeat

the majority principle, and, by the creation of small units, impede the employer in running his

plant (1950, 96).”

This authority meaningfully limited union activity, as is evident by the growing

hostility of the AFL to the Board after the passage of the Act. The authority to determine

28

bargaining units had the unintended effect of putting the NLRB in the center of the quickly

growing intra-labor fight between craft and industrial organizing. By all accounts, the decision

to give this power to the Board was made before anyone anticipated the split between the AFL

and the CIO in 1937. Yet by the AFL convention in that year, the AFL was so incensed by the

Board’s perceived favoring of industrial organizing that it unanimously adopted a resolution

to “assemble evidence in proof of the maladministration of the [Wagner] act,” to authorize the

AFL president and executive council to petition the president of the United States “for

prompt and adequate relief” and amendment of the Wagner Act (Gross 1974, 251). The AFL

would later join forces with opponents of the NLRA to support the Taft-Hartley amendments.

Section 9(c) gave the Board the right to oversee and certify the election of representatives,

while section 10(a) provided it with the right to prevent the unfair labor practices defined in

section 8.

The Board was thus empowered to issue legally enforceable orders for employers to

bargain with unions. The obverse is that this also empowered the Board to withhold

certification and bargaining orders when unions engaged in unacceptable behavior, violence in

particular. And indeed, the Board has on multiple occasions withheld bargaining orders from

otherwise entitled unions when they have been found to have engaged in severe violence

(Gitto 1982).7

On the labor side, the NLRA provided a set of benefits to unions and an institutional

structure to protect those benefits. But it also created a gatekeeper with the right and ability to

7 The Board has only rarely withheld bargaining orders, and the first time they did so appears to be in 1963, in the

Laura Modes Co. case, well after the union-side unfair labor practices were added. It is not clear whether the Board

could have done this prior to Taft-Hartley.

29

withhold those benefits for misbehavior. Unions that wanted access to the protections and

bargaining status provided by the NLRA thus had to take care not to antagonize the referee by

engaging in the type of violent unrest the Act was designed to prevent.

The industrial unions began to develop new tactics to shield them from accusations of

being perpetrators of violence. The most famous instance was the Flint sit-down strike in 1936-7

at General Motors. The workers locked themselves in and simply sat down, engaging in no work

while eschewing violence. The Flint Strike proved a pivotal moment (Bernstein 1969, 519-551).

The United Auto Workers (UAW) found a way to ensure that strikers could not be accused of

initiating violence against persons during a labor struggle; any violence would be initiated by the

employers or government. Moreover, the negotiations, led by John L. Lewis as president of the

new CIO, were not only with the company but involved, as well, the governor of Michigan and,

most importantly, Frances Perkins, the Secretary of Labor, who was throughout in close touch

with President Roosevelt. When the Governor of Michigan did consider enforcing an injunction

by calling in the National Guard, the Roosevelt administration prevented that move. This action

signaled the beginning of a new era, with the federal government intervening to enforce the rules

in employer-union conflict instead of siding with employers.

How did the NLRA come about?

The passage of the National Labor Relations Act of 1935 resulted from a remarkable

coalition that had not previously existed. Several factors made it possible.

The first was the growth of the labor movement. The numbers mobilizing in unions and

as voters started growing in the early 1930s. By 1939, in the aftermath of labor legislation, there

were more than 9 million union members (Katznelson 2006, 56). Union members were largely

30

democratic voters with union leadership deeply engaged in mobilizing their votes. Increasingly

the unions, especially those affiliated with the CIO, made demands of government to recognize

unions and their representation of workers, regulate labor conflict, and provide social insurance.

The CIO was key to a new coalition of Northern Democrats; they mobilized voters on their

behalf and persistently pressured for liberalizing legislation (Schickler forthcoming).

Lichtenstein documents labor demands for “industrial democracy;” for the workers “…the new

unionism represented not just a higher standard of living but a doorway that opened onto the

democratic promise of American life” (Lichtenstein 2002, 30).

The second factor stimulating change was fear and anxiety. The 1930s was “…an

anguish-filled environment… the most constant features of American political life continually

threatened to become unstable, if not unhinged“ (Katznelson 2013, 10).

There was fear of communism and revolution, and some federal officials and legislators came to

feel the labor unions and recognition of labor rights was a good bulwark against that threat

(Bernstein 1950, 102; Goldfield 1989, 1268-1269).

The other key to success of the legislation, well documented by Katznelson (2006, 53-67;

2013, 228-32) was the support of the Southern Democrats. A critical part of the New Deal

coalition, this faction of the Democratic Party went along with important labor legislation.

Unions were largely unheard of in the South, so this was a fairly easy trade of votes, but it came

at the price of the NLRA’s exclusion from the right to organize and bargain of occupations that

might attract African-Americans: the act explicitly exempted two of the biggest sources of

southern labor, agriculture and housework.

The changes in unions, the government, and business, during and partially as a

consequence of the Depression, combined to set the stage for a new equilibrium among labor,

31

employers, and government. The resulting compromise at once provided gains for each while

solving the problem of violence. The unions were willing to play by the rules and eschew

revolutionary aspirations and violence, but in return they expected union recognition, collective

bargaining, improved working conditions, and social benefits from both employers and

government. Business management – fearful of the disruptive effects of large-scale strikes and

worried that disorder and revolution were possible – became willing to accept terms with the

unions they previously rejected as the price of labor peace and productivity. The government

under the leadership of the Democratic Party came to recognize that it could gain electoral

support through union growth if it came to play a more impersonal, if still interventionist, role by

establishing a regulatory framework for labor-employer strife and enforcing the rules.

The NLRA was the final step in a series of efforts made in the wake of NIRA-inspired

unrest to improve and make permanent a set of institutions to encourage the peaceful resolution

of labor disputes. Section 1 of the NLRA makes clear the underlying assumption that legal

protection of the right to organize and bargain would increase safeguards on business by

increasing the capacity for “…the friendly adjustment of disputes.”

5. LABOR ORGANIZATION GAME REDUX

Period 2: 1933-36

The Great Depression represented a massive change in circumstances particularly for

workers who were unemployed or who feared losing their jobs. A surprisingly large portion of

workers were out of jobs for years, and the pro-labor movement grew. Moreover, as we

explained in section 3, sympathy for more radical change has grown among workers and within

32

organized labor. In addition, Democrats held a dominant party, whose support drew from labor

as well as business. These changes affect the preferences of the players, so that those in period 2

differ from those in period 1.

We model this change in circumstances by assuming that the probability of a successful

revolt has risen. During the Depression, violent aggression and revolt was more appealing that

prior (Katznelson 2013,10;449). In terms of the model, the danger of a revolt exceeded the

danger in the previous period. If the country failed to address labor problems, labor may well

choose radical action and defect from the Democratic Party. Thus we have that p2 > p1, where p2

is the probability of a successful revolt in period 2.

We do not analyze period 2 in detail. Instead, we focus on a particular comparative static

result reflecting the rise in probability of success if U chooses to revolt. We observe that there

exist a critical probability threshold, p*, such that if p2 > p*, U will choose revolt over striking.

We assume for this period that p2 is less than p*, but approaching it. Hence revolt and disorder

have become a real threat. In the absence of appropriate commitment technologies, this option is

not feasible. Outcome J, as discussed in the game in period 1 allows unions to take advantage of

firms.

The model also implies that, with the rise of p2 toward p*, the value of the outcome of

cooperation and negotiation between U and B has risen for all the players. During periods 1 and

2, commitment problems mean that this outcome cannot be implemented. Specifically, because

the Supreme Court has failed to give constitutional sanction to G’s labor regulation, G has no

way to enforce a set of impersonal rules governing union-business negotiation.

33

Period 3: The NLRB Comes on Line

The final period we study begins in early 1937 with the Supreme Court’s ruling that the

NLRA and its progeny, the NLRB, were constitutional, allowing the NLRB to enforce its rules.

Circumstances in period 3 differ from period 2 in several ways. Because of the threat of disorder,

many businesses came to favor compromise, as represented in the NLRA, which gained far more

than majority support in Congress. For our purposes, the main implication of this ruling is that

the government gained the ability to sanction both labor and business if they fail to play by the

(NLRA/NLRB) rules. As we show, the ability of G to sanction both business and labor for

failing to play by the rules allows the three players to implement cooperation between U and B

that involves mutual respect for the rules and hence an absence on violence.

To see how this cooperative equilibrium works, we modify the game to reflect G’s

enforcing the NLRB rules. As before, U has three choices at the first node of the game, though

they differ somewhat from U’s choices in period 1. First, U may choose to play by the rules;

second, it can continue striking without limits, hence retaining violence potential; and third, it

can choose to revolt. We assume that the probability of a successful revolt (from U’s standpoint)

is p2, close to p*, but not quite; as before, we assume that if the Depression continues and labor

sees too few concessions, p2 may rise above p* (see figure 2). As in game 1, if U chooses to

revolt, then nature chooses whether labor is successful (outcome W) or if disorder results

(outcome X). The implied lottery is L33 = p2W + (1 – p2)X.

If U chooses to play by the rules or to continue striking without limits, then B has the

next move and may choose from among 3 options: fight (continue to use violence against U),

contest U within the system, or to play by the rules, here meaning recognize the union and

34

negotiate with it in good faith. Finally, if B chooses to fight or contest U, then G, now in the form

of the NLRB, must choose whether to enforce the rules.

We make the following assumptions about the players’ preferences. U prefers the result

when it plays by the rules and B chooses to also play by the rules (outcome O) to two other

outcomes: (i) the outcome of revolting (L33 = p2W + (1 – p2)X); and (ii) the outcome when it

chooses continuation of the status quo (striking without limits, B contests U’s actions, and then

suffer punishment by G).

For B, we assume that if U chooses to play by the rules, B prefers to play by the rules. If

U chooses to continue striking without limits, it prefers to contest U’s actions. For G, reflecting

its expectation of electoral gain by having solved the problem of labor violence, we assume that

it prefers enforcing the rules over doing nothing. In terms of the model, this means that G will

punish either party if it fails to play by the rules. As with the failed NIRA, if G fails to implement

the rules, it will gain no electoral support from the new labor legislation as the violence between

labor and business continues, possibly allowing p2 to rise above p*, leading U to choose revolt

instead of cooperation.

As before, we solve the game using backward induction. At each decision node involving

G, G will choose to enforce the rules. Doing so gains it electoral support and forestalls potential

revolution. Working back a node, B has three choices: if U has chosen to play by the rules,

fighting risks punitive actions by G so B is best off choosing to play by the rules. If, instead, U

chooses to continue striking without limits, then B is best off contesting U’s actions and G will

impose sanctions on U. We now come to the first choice of the game, U’s decision among

playing by the rules, striking without limits, and revolting. If U chooses to play by the rules,

outcome O occurs in which both U and B play by the rules and hence negotiate in good faith. In

35

contrast, if U chooses to continue to strike without limits, then B will contest U’s strikes and G

will then rule against U. Where as if U chooses to revolt, the outcome is that of the lottery L33 =

p2W + (1 – p2)X. Among these three options, U prefers to play by the rules. Hence the

equilibrium path of this game is for U to choose to play by the rules and B to choose to play by

the rules.

The main conclusion is that the new mechanisms created by the NLRA create G as an

impersonal enforcer of the new rules. The Act provides for meaningful sanctions on either B or

U if they violate the rules; in particular, if either chooses violence. By creating the appropriate

structure and process, the Act created the appropriate commitment mechanisms, allowing the

three parties to sustain a new, cooperative equilibrium.

6. A Dialogic Reinterpretation of the Constitutional

Controversies over the New Deal

The NLRA was the culmination of several decades of legal innovation, innovation that is

largely responsible for contemporary administrative law jurisprudence. Politics were an obvious

component of the eventual finding of New Deal laws as constitutional beginning in 1937. But the

traditional account of the New Deal Constitutional controversies over-emphasizes politics and

under-emphasizes the role of the development of doctrine and the mechanisms of administrative

delegation. The standard wisdom is that after FDR threatened to “pack the court,” Justice

Roberts made his famous “switch in time,” and the Justices acquiesced to his New Deal

legislation. Although a caricature, this brief summary of the standard wisdom in constitutional

law books captures their essence.

36

We argue that a far more complex and interesting story hides in legal doctrine (Cushman

1998 makes a similar claim). The NLRA was a clear and direct attempt to respond to concerns

about the New Deal’s constitutionality as articulated by the Court in the early New Deal cases.

By doing so, Congress invented new structures and processes that the Court would hold in

National Labor Relations Board v. Jones & Laughlin Steel Corporation (301 U.S. 1, 1937) as

satisfying constitutional restrictions.

We assert that Congress and the Court engaged in a dialogue concerning issues of

delegation, political control, oversight, and the means of ensuring rights of due process. By

trying new structures and processes and having them, at times, struck down and, at times, upheld,

Congress and the Court jointly created a major expansion of administrative law. And, in the

process, they were able to ensure enforcement of the law and the credibility of its commitments.

In the two precedents most relevant to the NLRA, Panama Refining and Schechter, the

Supreme Court analyzed two constitutional issues with great care: 1) the structure and process of

regulatory delegation; and 2) Congress’ regulatory authority under the commerce clause. In these

precedents, we see Congress, the President, and the Court struggling to interpret the commerce

clause and to define the bounds of proper regulatory delegation.

Both Panama Refining and Schechter were decided prior to the passing of the NLRA in

Congress. While Panama Refining was decided on January 7, 1935, prior to Congress’

consideration and passage of the Wagner Act; Schechter was decided on May 27, 1935, between

the dates of the Senate and House debates of the Act. And while they were not the only

precedents that the drafters of the NLRA had to contend with, they were representative of the

general issues plaguing the New Deal Acts.

37

So as to better understand the dialogic nature of the relationship between the Court and

Congress in constitutional cases, we take a closer look at these cases and how Congress

responded to them in the NLRA. Lastly, we will look at how the Court responded to Congress’

implementation of its guidelines through its decision in Jones & Laughlin Steel.

A. The Delegation Issue

Panama Refining Panama Refining Co. v. Ryan (293 U.S. 388,1935) was the

constitutional challenge to the regulation of petroleum goods under the National Industrial

Recovery Act (NIRA). The Court held this regulation an improper and hence unconstitutional

delegation of legislative power (431). In Panama Refining, the Court side-stepped the commerce

clause issue as a majority of the Justices agreed to the unconstitutionality of the regulation based

on the delegation question.

Importantly, the Court did not simply rule Congress' actions as outside the bounds of

constitutionality. Through a careful analysis of previous instances of delegation why that it had

approved delegation and why this case presented delegation that failed to meet these criteria.

The NIRA incorporated many provisions now considered outlandish in their failure to

assure due process and to prevent arbitrary and capricious public actions. As Irons (1983) and

others suggest, the authors of this legislation paid little attention to constitutional issues;

specifically, they made little effort to assure that the legislation would meet the standards set in

recent cases approving regulatory delegations by the national government. For example, they: 1)

delegated authority without sufficient definition of terms or limits on authority; 2) delegated

regulatory authority to private groups; 3) paid little attention to legal decisions about existing

legislation with which the New Deal legislation interacted (such as the Federal Trade Act); and

38

4) did not ensure respect for rights of due process; for example, it did not require that the

president make a finding prior to acting.

The Court pointed to these delegation failures of the NIRA in Panama Refining (415). It

then argued that, to approve this new set of regulations, Congress needed to set clear guidelines

and procedures to the agency.

Schechter also raised the delegation issue. In A.L.A. Schechter Poultry Corp. v. United

States (295 U.S. 495, 1935), a unanimous Court ruled that the Act was unconstitutional as it was

both an invalid delegation of legislative power and an improper regulation of interstate

transactions with only an indirect effect upon intrastate commerce.

In Schechter, as in Panama Refining, the Court was critical of Congress’ haphazard

delegation of power (541-2). However, the Court went further in Schechter, not only describing

the failures on the NIRA delegation scheme but also comparing it to a proper administrative

delegation scheme, that of the Federal Trade Commission. In Schechter, the Court compared the

administrative delegation in the Federal Trade Commission Act (FTCA) and the NIRA and

found that, while the FTCA contained adequate safeguards and limits, the NIRA lacked them

(534-535).

If one of the main failures of the drafters of the NIRA was that they paid little

attention to legal decisions about existing New Deal legislation, the drafters of the NLRA

proved to be not only attentive to preceding legal decisions but also directly responsive to

them. From the very beginning the proponents of the Wagner Act, as the NLRA became

known, were “[C]ommitted to the model of a full blown, full fledged judicial agency like

the Federal Trade Commission.” (Bernstein 1950, 228). This is a model that had been

39

previously accepted by the Court (see Federal Trade Commission v. Gratz, 253 U.S. 421,

1920).

Further, unlike the drafters of the NIRA, drafters of the Wagner Act were

meticulous in delineating the agency’s power. For example, Section 10 of Act provides

that: “[T]he Board is empowered, as hereinafter provided, to prevent any person from

engaging in any unfair labor affecting commerce.” Every part of this declaration is

defined and limited. The “Board,” “person,” “unfair labor practice,” “affecting

commerce,” and “commerce” are all defined in the Act, thus delineating under which

circumstances and over whom the administrative agency has authority. Furthermore, the

NLRA delegated authority to an administrative agency, not private industry boards.

This attention to the New Deal precedent and concerted effort to address the Court’s

concerns paid off. In Jones & Laughlin Steel, holding the Wagner Act constitutional, the Court

acknowledged that Congress had fixed the delegation issue under the NIRA. After declaring that

the Schechter case is “not controlling here,” (41) the Court goes on and outlined the standards to

which the Board had to conform (47) and declared that the Act properly defines and delineates

the scope of the Board’s authority (30).

The NIRA had: 1) delegated authority without sufficient definition of terms or limits on

authority; 2) delegated regulatory authority to private groups; 3) paid little attention to legal

decisions about existing legislation with which the New Deal legislation interacted; and 4) did

little to ensure respect for rights of due process. Therefore, the Court ruled it unconstitutional.

The Wagner Act deliberately sought to remedy these defects; it 1) delegated authority with

sufficient definitions of terms and limits on authority; 2) delegated authority to the National

Labor Relations Board, a government administrative agency; 3) responded to concerns expressed

40

by the Court in previous New Deal cases and modeled the administrative schema on an existing

and established agency; and 4) ensured due process rights through delineating the structure and

processes through which the agency was to exercise its authority. It learned from the Court’s

previous decisions and when drafting the NLRA, Wagner and his writers placed the new agency

comfortable within constitutional bounds.

B. Congress’ Regulatory Authority under the Commerce Clause

In the delegation issue, the Court communicated clear criteria to Congress. It explained in

both Schechter and Panama Refining proper and improper delegation of power. The dialogue

between the Court and the President and Congress on the issue of the commerce clause was less

clear. For several decades, the Court struggled the nature constitutionally proper bounds of

federal power; at times sending conflicting messages to Congress. However, as in the delegation

cases, the Court and Congress engaged in a bargaining process over the national government’s

commerce authority that ultimately determined the constitutional bounds.

The Dialogue between Court and Elected Branches, 1890 to 1932

The dialogue between the Court and the elected branches over the commerce clause had

always been present. With respect to the issue of labor and labor violence, it picked up in 1895.

At that time, Congress sought to exercise its commerce clause power in the Sherman Act (15

U.S.C. §§ 1-7, 1890). In 1895, the Court applied the Sherman Act to labor struggles. In In re

Debs (158 U.S. 564,1895), the Court tested and upheld the validity, under the commerce clause,

of the labor injunction to stop strikes in the railways. After the Pullman strike ended, Debs and

six others were charged with contempt of court for violating a labor injunction. They perpetrated

no violence but were found guilty of communicating with strikers. For that, they were sentenced

41

to 6 months in prison. The Court upheld the prison sentence holding that it was proper for the

Sherman Act to apply to labor injunctions as railway labor struggles impeded the interstate

movement of goods. Thus, these struggles were under Congress’ commerce clause power. In re

Debs was a broad reading of federal powers and expansion of the application of the Sherman

Act.

This decision followed a narrow reading of the Sherman Act in United States v. E. C.

Knight Co (156 U.S. 1, 1895). The Supreme Court decided this case just five months before In re

Debs. In E.C. Knight, the Court interpreted the Sherman Act narrowly, ruling that the Sherman

Act did not cover manufacturing as it only indirectly affected interstate commerce. Thus,

manufacturing was beyond Congress’ commerce clause power, which covered only direct effects

to interstate commerce.

The Court's indecision on how broadly to define the commerce clause led to many years

of inconsistent jurisprudence. Between E.C. Knight and the New Deal cases, the Court tried

several different tests to determine the bounds of the commerce clause. For example, the Court’s

ruling on constitutionality in E.C. Knight hinged on the test of direct vs. indirect effects on

interstate commerce. However, in Stafford v. Wallace (258 U.S. 495,1922), the Court broadened

the test and argued that the entire “stream of commerce among states” was under the regulatory

power of the commerce clause. It explained that, “streams of commerce among the states are

under the national protection and regulation, including subordinate activities and facilities which

are essential to such movements, though not of interstate character when viewed apart from them

(519).” The Court’s struggles about the narrowness of the commerce clause arose in response to

Congress’ attempts to use the commerce clause to regulate different aspects of commerce. As

Congress tried new regulatory structure and process in various regulatory legislation, the Court

42

upheld or struck down these new mechanisms, thus continually refining the complex definition

of the commerce clause.

While In re Debs applied the Sherman Act to railway strikes, an inherently interstate

venture, Loewe v. Lawlor (1908) broadened the Act’s application to labor. Loewe involved the

United Hatters of North America's charges of restraint of trade under the Sherman Act due to

their boycott activities. The workers were not involved in interstate commerce and did not

obstruct the movement of goods, as had happened In re Debs. Even so, the Court concluded that

their acts must be considered as a whole and thus found that the purpose of their conspiracy was

to prevent manufacture and inhibit interstate commerce. In yet another reading of the commerce

clause after E.C. Knight, in Loewe, the Court found that the Sherman Act does apply to

manufacturing as it concerns labor. The loss in Loewe led labor unions to demand legislative

changes. In particular, they wanted to be excluded from coverage under the Sherman Act. Labor

unions obtained an attempt to create this exclusion with Section 6 of the Clayton Act (15 U.S.C.

§ 12-27, 1914). However, in Duplex Printing Press Co. v. Deering (254 U.S. 443, 1921), the

Court reaffirmed its prior ruling in Loewe by rejecting a blanket exclusion of labor from antitrust

laws.

The NLRA

The writers of the NLRA drafted the Act in the shadow of this inconsistent commerce

clause jurisprudence. In his legislative history of the Wagner Act, Bernstein explains that even as

the drafters worked on perfecting “the bill’s substantive and procedural provisions,” they were

aware of the “constitutional quicksand on which they rested” (229). Ultimately, regardless of the

delegation issue, the drafters knew that “the Wagner Act was bottomed on the commerce

43

clause” (229). In an effort to prompt the Supreme Court to give an expansive reading to the

clause, the drafters heavily borrowed legal language from the Court’s decisions in favorable

commerce clause cases.

In the bill’s carefully crafted Declaration of Policy, the Act’s drafters made sure to argue

that the denial of the rights to organize and bargain collectively necessarily led “to strikes and

other forms of industrial strife and unrest,” which in turn constricted the flow of goods into the

“channels of commerce” and adversely affected level of employment and wages” (229-230). The

drafters took this language directly from Gibbons v. Ogden (22 U.S. 1, 1824) and Stafford v.

Wallace (1922), two of the Supreme Court’s broader readings of the commerce clause.

While drafts of the NLRA were being circulated in Congress, Schechter was decided.

Although the Court abstained from the commerce clause issue in Panama Refining, it engaged

with it directly in Schechter, finding that Congress’ power under the commerce clause did not

reach into what the Court held to be intrastate matters. It explained that, “the authority of the

federal government may not be pushed to such an extreme as to destroy the distinction, which

the commerce clause itself establishes, between commerce ‘among the several States' and the

internal concerns of a state... We are of the opinion that the attempt through the provisions of the

code to fix the hours and wages of employees of defendants in their intrastate business was not a

valid exercise of federal power” (549-50).

Concerned about the Court’s most recent narrow reading of the commerce clause, the

drafters of the NLRA immediately changed the language of the act so as to accommodate the

Court’s objections in Schechter. In response to Schechter, Wagner had the Declaration of Policy

rewritten before the House debate. “It was revised, to emphasize the effect of labor disputes on

interstate commerce and to de-emphasize the mere economic effects which had been rejected by

44

the Court… For example, the bill’s definition of the term ‘affecting commerce’ was changed

from acts ‘burdening or affecting commerce’ to those ‘burdening or obstructing

commerce.’”(n.16, internal citations omitted).

Their efforts of the Act’s drafters paid off. The NLRA was challenged in National Labor

Relations Board v. Jones & Laughlin Steel Corporation (301 U.S. 1, 1937). In Jones & Laughlin

Steel, the Court held that the Act’s purpose was proper under the commerce clause (31).

Wagner’s borrowing of favorable commerce clause cases and changes to the Declaration

of Policy appeared to work. The language used – language that the Court had previously found to

be constitutional – increased the likelihood that the NLRA would be upheld. In Jones &

Laughlin Steel, the Court settled on the bounds of constitutionality as it related to the commerce

clause. Following, Jones & Laughlin Steel, the Court made a concerted effort to clean up its

inconsistent commerce clause jurisprudence.

The Court's inconsistency and Congress' efforts to pass new and creative laws are

different sides of a dynamic dialogue and bargaining process around constitutional issues. At the

same time that the Court struggled to determine the new bounds of federal power, Congress

struggled to figure out what new forms of that power would survive constitutional muster.

Instead of seeing the Court as stubbornly obtuse and only willing to change once

threatened, it is more advantageous to see it as a conservative body trying to keep order in times

of great uncertainty. Rather than interpret the 1937 “switch” as a strictly political move by a

bullied Court, we should view it as part of the dialogue over the content of new constitutional

doctrines in which Congress and the president accommodated many of the Court’s concerns,

especially with respect to the delegation of political authority to regulatory agencies.

Reciprocally, the Court accommodated the political branches’ authority to address national crises

45

through the delegation of regulatory authority. The Court, the President and Congress tried,

rejected and accepted various new configurations of political control and oversight. By the end

of the 1930s and the early 1940s, they have worked out the new technology of delegation. This

new technology of delegation is one solution to the commitment problems in the labor violence

case.

7. Implications

The games modeling the interaction of labor, business, and government reveal important

insights about labor history, violence, and, more generally, about open access. These games

allow us to interrogate the history to answer all three major questions asked at the outset: why

did labor violence persist for so long; how did the NLRA finally solve the problem of violence;

and why did the government have incentives to design the NLRA and to become a impersonal

player in the NLRB regulatory process?

The model shows why the violence associated with labor persisted for so long. We

stipulate that both labor and business would be better off under a regulatory framework that

fostered cooperation rather than confrontation. In the absence of a impersonal government,

neither U nor B could commit to foreswear violence. Each side could take advantage of the other

by continuing to use violence even if the other has stopped using violence. For this reason the

political equilibrium involved a century of violence and a lack of cooperation.

The design of the NLRA, in combination with the Supreme Court’s sanction of the law,

altered both the set of moves available to the players and their incentives. This legislation gave

the government the ability to behave as an impersonal party overseeing the rules involving

recognition of unions and to punish both parties for violating the rules.

46

The key to the new, post NLRA cooperative equilibrium is threefold. First the on-going

and lengthening Depression increased the likelihood of success of a revolt by labor, making this

action more plausible. This outcome would make both B and G much worse off. The potential

attractiveness of revolt to U, in turn, raised the value of compromise to B and G; for this reason

they were willing to help design a new system. Third, the new equilibrium involved G inventing

a new regulatory process, which included the ability to sanction both firms and labor unions for

violating the rules. Democrats in control of the national government gained electorally from a

law that at once solved the problem of violence and allowed it to become a impersonal player

administering the rules, enforcing the rules against either party in the event that they break the

rules.

This logic also explains why the NLRA solved the century-long problem of violence

where previous administrations could not. The threat of revolt was not present. Further, the

Republican Party dominated the national government from 1880-1930 and did not believe it

could gain electorally by legalizing unions and creating an impersonal regulatory process

governing the labor process. In the wake of the Great Depression, Democrats took control of the

national government in 1933; overtime, they came to see that they could benefit from helping

labor to organize.

We realize that this paper is just one step in the analysis of open access for labor. Major

aspects of labor behavior, regulation, and open access would appear during WWII and the

immediate years following the war, including the 1947 Taft-Hartley legislation that reduced the

perceived bias of the system in favor of the labor.

We conclude by discussing the larger issues in the construction of open access. Open

access to labor organizations seemed impossible as long as the commitment problems remained

47

unsolved and the violence problems continued. Our approach highlights problems with open

access faced by other civil society organizations, such as the civil rights groups and the women’s

movements. In these and similar cases, the demands for recognition threatened powerful

interests who had the political and economic clout to inhibit open access. The American

experience with open access to business corporations therefore occurred decades earlier than

open access to labor organization.

Our approach has larger implications for political development and open access. By

investigating the violence problem as an important factor in determining when certain kinds of

organizations and groups attain open access, we are better able to address the uneven

development of open access and its consequences for pluralist democracy. Open access to the

corporate form of organization in the United States occurred in the 1840s. North, Wallis,

Weingast (2009) note the importance of this access for economic and political development.

Similarly, Brooks and Guinnane argue (this volume), “acquisition of corporate rights was key to

the growth and success of civil society.” The first set of authors miss the limited and fragile

access of other organizations and voices that define a pluralist democracy. The second recognize

that open access for some organizations is hard-won, but they ignore the importance of the

violence problem.

The perspective of this paper has implications for several of the other papers studying

open access to business organizations. Until recently few worried about the timing or explanation

of these, as if the value of the corporate form itself was sufficient explanation for its existence.

The common view has no explanation for why new business forms emerged, the timing of that

emergence, or the administrative apparatus designed to support open access to these business

forms.

48

Many questions remain. Why do we observe so much experimentation and hesitation

with the legislation proposing to open access for firms? This hesitancy and experimentation

suggest that the proponents of the new legislation were trying to solve problems – what were

these? Put another way, the opening of access to the corporate form seems to have been

characterized by problems whose solution was not obvious in the beginning. Moreover, the

general incorporation (see Lu and Wallis this volume) that opened access for other organizations

was not necessarily the best route to open access for labor organizations, whose essential

associational rights comprise not only legal recognition but also collective bargaining and the

right to strike.

Finally, our paper has implications for the maintenance of democracy and open access

over the long-term. Even the strongest and most stable of democracies face episodic threats to

their survival; the mystery is how they manage to survive these threats and whether they can

continue to do so.8

Democratic stability cannot be taken for granted. As the problem of labor

violence illustrates, nothing made peaceful resolution inevitable; and, under other condition (e.g.,

absent the Great Depression), the long-term violence equilibrium might have continued longer.

Open access, a presumed hallmark of democracy, is part of the solution to maintaining

democratic stability, but it also creates part of the problem. Open access is an ideal type. In

practice, those countries approximating open access orders moved piecemeal toward greater

openness, meaning that they allowed open access for some types of groups and organizations

while suppressing others. The uneven spread of open access means the suppression of certain

8 Mittal (2011) and Mittal and Weingast (2013) study the process of adaptive efficiency in the face of

changing circumstances; that is, the degree to which countries adapt peacefully to major shocks in their

environment.

49

voices holds the risk of disruption into protest, violence, and disorder.

The century-long history of labor violence is not unique in American history. Southern

suppression of African-Americans and the U.S. suppression of Native Americans represent other

cases. Illegal, violent groups, such as the Molly Maquires or the Klu Klux Klan, also occur with

some frequency. Maintaining democratic stability requires a constant process of balancing

existing interests, accommodating new or previously excluded interests, and suppressing groups

who use violence to accomplish their ends. This process of accommodation and change

necessarily implies risk. Success at each stage is not assured, so democracy and open access

remain fragile to a degree, even in the seemingly most stable of countries.

At issue in this paper is how failures of American pluralist politics contribute to the

fragility of U.S. democracy. In Ira Katznelson’s magisterial Fear Itself (2013), he demonstrates

how the New Deal saved a floundering democratic state that the Depression had destabilized.

The emergence of a procedural state might appear to offer favorable conditions for an open

access society, but in fact it also produces the possibility for the blocking of organizations by

powerful interests within Congress. The new institutional arrangements permitted the serious

post-WWII violations of privacy and liberty embodied in the practices of the House Un-

American Activities Committee, Senator Joseph McCarthy, and the FBI as they sought out

communists and communist sympathizers.

These observations raise the deeper question of what are the requirements for a truly open

access society? When does government behavior violate open access and thus threaten the

equilibrium open access helps bulwark?

50

UB

N

G

Revolt

Strike without limits

Fight

Strike within limits

p1

q2Side w B

Do nothing

Negotiate

Figure 1:1880-1930.

Large benefits to U

Disorder

N

B

G

Fightq1

Side w B

Do nothing

Negotiate

N

OutcomesD

E

I

J

D

E

F

H

A

C

51

UB

N

Revolt

Strike without limits

Play by the rules

p2

Figure 2:1937 forward.

Large benefits to U

Disorder

B

Fight

Play by the rules

OutcomesK

LM

NOR

ST

UV

W

X

enforce rules

Do nothingG

enforce rules

Do nothing

G

contest

Fight

enforce rulesDo nothing

G

enforce rules

Do nothing

G

contest

Play by the rules

52

References

Adler, William M. 2011. The man who never died: the life, times, and legacy of Joe Hill,

American labor icon. New York: Bloomsbury.

Ahlquist, John S., and Margaret Levi. 2013. In the Interests of Others. Princeton: Princeton

University Press.

Bachrach, Peter, and Morton S. Baratz. 1962. "Two Faces of Power." American Political Science

Review 56 (4):947-52.

———. 1963. "Decisions and Nondecisions: An Analytical Framework." American Political

Science Review 57 (3):632-42.

Bernstein, Irving. 1950. The New Deal Collective Bargaining Policy. Berkeley: University of

California.

———. 1969. The lean years; a history of the American worker, 1920-1933. Boston: Houghton

Mifflin.

———. 1971. Turbulent years; a history of the American worker, 1933-1941. Boston: Houghton

Mifflin.

———. 1985. A caring society: the New Deal, the worker, and the Great Depression : a history

of the American worker, 1933-1941. Boston: Houghton Mifflin.

Brecher, Jeremy. 1997. Strike! Cambridge, MA: South End Press.

Cox, Gary W., Douglas C. North, and Barry R. Weingast. 2014. "The Violence Trap: A Political-

Economic Approach to the Problems of Development." Hoover Institution, Stanford University.

Currie, Janet, and Joseph Ferrie. 2000. "The Law and Labor Strife in the United States, 1881-

1894." The Journal of Economic History 60 (1):42-66.

Dahl, Robert A. 1961. Who governs? Democracy and power in an American city. New Haven:

Yale University Press.

Edwards, P.K. 1981. Strikes in the United States, 1881-1974, Oxford: Basil Blackwell.

Eidlin, Barry. 2009. "'Upon this (foundering) rock': Minneapolis Teamsters and the

transformation of US business unionism, 1934-1941." Labor History 50 (3):249-67.

Foner, Philip S. 1947 (1988). History of the Labor Movement in the United States, vol. 1: From

Colonial Times to the Founding fo the American Federation of Labor. 1988 ed. New

York: International Publishers.

Forbath, William E. (1991) Law and the Shaping of the American Labor Movement. Cambridge:

Harvard University Press.

Franfurter, Felix, and Nathan Greene. 1930 [1963]. The Labor Injunction. Gloucester, Mass.:

Peter Smith

Friedman, Gerald. 1988. "Strike Success and Union Ideology: The United States and France,

1880-1914." The Journal of Economic History 48 (1):1-25.

Geraghty, Thomas M., and Thomas Wiseman. 2011. "Conflict and compromise: Changes in U.S.

strike outcomes, 1880 to 1945." YEXEH Explorations in Economic History 48 (4):519-

37.

Gitelman, H. M. 1973. “Perspectives on American Industrial Violence” Business History Review,

47(1): 1-23

Gitto, Donald R. 1982. “Strike Violence: The NLRB’s Reluctance to Wield its Broad Remedial

Power,” Fordham Law Review 50(6): 1371-1398.

53

Golden, Miriam. 1997. Heroic Defeats: The Problems of Job Loss. Cambridge: Cambridge

University Press.

Goldfield, Michael. 1989. "Worker Insurgency, Radical Organization, and New Deal Labor

Legislation." American Political Science Review 83 (4):1257-82.

Gross, James A. 1974 The Making of the National labor Relations Board: A Study in Economics,

Politics, and the Law. Volume 1 (1933-1937). Albany, NY: State University of New

York Press.

Hattam, Victoria C. 1993. Labor visions and state power : the origins of business unionism in the

United States. Princeton: Princeton Univ. Press.

Irons, Peter. The New Deal Lawyers. Princeton: Princeton University Press, 1982.

Johnson, Victoria. 2008. How Many Machine Guns Does it Take to Cook One Meal? The Seattle

and San Francisco General Strikes. Seattle: University of Washington Press.

Katznelson, Ira. 2006. When Affirmative Action Was White. New York: W. W. Norton.

———. 2013. Fear Itself: The New Deal and the Origins of Our Time. New York: Liveright

Pub. Corp.

Levi, Margaret. 1977. Bureaucratic Insurgency: The Case of Police Unions. Lexington, MA:

Lexington Books.

Lichtenstein, Nelson. 2002. State of the Union. Princeton: Princeton Univ. Press.

Lukas, J. Anthony. 1997. Big Trouble. New York: Simon & Schuster.

Lukes, Steven. 2005 [1974]. Power: A Radical View. New York: New York : Palgrave

Macmillan.

Mares, Isabela. 2003. The Politics of Social Risk. New York: Cambridge University Press.

North, Douglass C., John Wallis, and Barry Weingast. 2009. Violence and Social Orders: A Conceptual

Framework for Interpreting Recorded Human History. New York: Cambridge University Press.

Perlman, Selig. 1928. A theory of the labor movement. New York: Augustus M. Kelley.

Rogin, Michael. 1962. "Voluntarism: The Political Functions of an Anti-political Doctrine."

Industrial and Labor Relations Review 15 (4):521–35.

Schickler, Eric. Forthcoming. The Civil Rights Realignment. Princeton: Princeton University

Press.

Swenson, Peter. 2002. Capitalists Against Markets. New York: Oxford Paperbacks.

Taft, Philip, and Philip Ross. "American labor violence: Its causes, character, and

outcome." Violence in America: Historical and comparative perspectives 1 (1969): 221-

301.

Tomlins, Christopher L. 1985. The State and the Unions: Labor Relations, Law and the

Organized Labor Movement in America,1880-1960. New York: Cambridge University

Press.

Voss, Kim. 1993. The making of American exceptionalism: the Knights of Labor and class

formation in the nineteenth century. Ithaca: Cornell University Press.

Cases Cited

Gibbons v. Ogden, 22 U.S. 1 (1824)

United States v. E. C. Knight Co, 156 U.S. 1 (1895)

In re Debs, 158 U.S. 564 (1895)

Loewe v. Lawlor (1908)

Federal Trade Commission v. Gratz, 253 U.S. 421 (1920)

54

Duplex Printing Press Co. v. Deering, 254 U.S. 443 (1921)

Stafford v. Wallace, 258 U.S. 495 (1922)

Panama Refining Panama Refining Co. v. Ryan, 293 U.S. 388 (1935)

A.L.A. Schechter Poultry Corp. v. United States, 295 U.S. 495 (1935)

National Labor Relations Board v. Jones & Laughlin Steel Corporation, 301 U.S. 1,

(1937)

United States v. Darby Lumber Co., 312 U.S. 100 (1941)

Laws Cited

Sherman Act (15 U.S.C. §§ 1-7, 1890)

Federal Trade Commission Act (15 U.S.C §§ 41-58, 1914)

Clayton Act (15 U.S.C. §§ 12-27, 1914)